Hybrid Electric Vehicles (Principles and Applications with Practical Perspectives) || Batteries, Ultracapacitors, Fuel Cells, and Controls

11Batteries, Ultracapacitors, FuelCells, and Controls

11.1 Introduction

In this chapter, the requisite energy storage systems for EVs and HEVs are discussed.In a HEV or PHEV, onboard batteries or ultracapacitors are charged from the internalcombustion engine/generator set or from the electric power grid. The chemical energystored in batteries is converted to electrical energy for traction motor and vehicle propul-sion. Also, energy storage systems are responsible for recuperating regenerative brakingenergy to further increase vehicle efficiency. Thus, the performance of EVs and HEVsdepends on energy storage systems to a large extent. Therefore, this chapter is devotedto a discussion of battery, ultracapacitor, and fuel cell technology. Here, we will focuson the techniques of modeling, hybridization, equalization, charging control for batteriesand ultracapacitors.

Batteries are made of cells where chemical energy is converted to electrical energyand vice versa. The battery energy storage system (BESS) comprises mainly batteries, thepower electronics-based conditioning system, and a control system. In HEVs, batteriesprovide energy for the traction motor and store regenerative energy; a power electronicsconverter, typically of bidirectional capability, provides an interface between the batteriesand power produced by the onboard internal combustion engine or utility power in thecase of a PHEV; the control system is responsible for power and energy managementincluding charging/discharging and equalization control. The above description applies aswell for the ultracapacitor energy storage system (UESS). The main difference is that thebattery is an electrochemical energy conversion device while the ultracapacitor does notinvolve any chemical reactions. The BESS or UESS topology is illustrated in Figure 11.1.

In order to have the desired voltage rating and current rating for application in HEVs,many cells must be connected in series and/or in parallel in the BESS or UESS. Voltagebalancing or equalization is required if more than three cells are connected in series.

Generally speaking, a battery has the characteristics of high energy density andrelatively low power density. The internal resistance is the major factor for its limiteddischarging and charging current capability. The internal equivalent series resistance

Hybrid Electric Vehicles: Principles and Applications with Practical Perspectives, First Edition.Chris Mi, M. Abul Masrur and David Wenzhong Gao.© 2011 John Wiley & Sons, Ltd. Published 2011 by John Wiley & Sons, Ltd. ISBN: 978-0-470-74773-5

316 Hybrid Electric Vehicles

BESSor

UESS

BidrectionalDC/DC

ConverterDCBus

Control

Figure 11.1 HEV BESS/UESS topology

(ESR) has different values under charging and discharging operating conditions. Thevalues are also dependent on the frequency of the discharging current [1]. For lithium-ionbatteries, the internal resistance could increase by 50% from 1000 Hz to 100 Hz. Theampere-hour capacity is affected by the discharging current rate and is modeled byPeukert’s equation [2], C p = I k t , where k is the Peukert constant; k = 1 for an idealbattery. The charging and discharging efficiencies are nonlinear functions of current andthe state of charge (SOC).

The battery can be modeled as an equivalent circuit such as an internal resistancemodel or a resistance–capacitance (RC) model in ADVISOR [3]. In an internal resistancemodel, a battery is modeled as a voltage source and an internal resistor. Both the voltagesource and the internal resistor are functions of the SOC and temperature, which canbe implemented as lookup tables. In a RC model, a battery is represented as a parallelcombination of two RC branches. The very large capacitor models the battery’s chargecapacity while the smaller capacitor models the time constant due to surface effects thatlimit the current. The model can be implemented as an S function in MATLAB/Simulink.

On the other hand, when compared to common capacitors, the ultracapacitor (electricdouble-layer capacitor) has a very high energy density, which could be thousands of timesgreater than a high-capacity electrolytic capacitor. Larger double-layer capacitors can havecapacitances up to 5000 F as of 2010. Compared to batteries, an ultracapacitor has thecharacteristics of high power density and relatively lower energy density. Its equivalentinternal resistance is decades lower than that of a battery, thus allowing decades of higherdischarging/charging current. Its overall round-trip efficiency is higher than that of abattery. Its capacitance is huge compared to an ordinary electrolyte capacitor, allowingenough energy storage for HEV acceleration power requirements. Note that the internalresistance and capacitance are highly dependent on the frequency because of the porousnature of the electrodes. One big advantage of the ultracapacitor is that its SOC is allowedto vary more widely and thus has longer life cycles. Its capability to provide high powerbursts is ideal for hybrid vehicle applications.

An ultracapacitor can also be modeled as an internal resistance model or RC model inthe same way as for a battery. The difference is that the ultracapacitor’s internal resistancefor charging is typically the same as for discharging.

To predict the behavior of battery/ultracapacitor voltage and current during transientoperation such as acceleration and deceleration, physics-based dynamic models are neededto account for the time constants due to the electrochemical reactions in batteries ordouble-layer effects in ultracapacitors.

Batteries, Ultracapacitors, Fuel Cells, and Controls 317

11.2 Battery Characterization

Capacity (C)

The battery capacity specifies the amount of electric charge a battery can supply before itis fully discharged. The SI unit of battery capacity is the coulomb. A more general unitfor battery capacity is ampere-hour (Ah), with 1 Ah = 3600 C. For example, a battery of20 Ah can supply 1 A current for 20 hours or 2 A for 10 hours, or in theory 20 A for1 hour. But in general, the battery capacity is dependent on discharge rate.

There are two ways of indicating battery discharge rate: C rate is the rate in amperes,while nC rate will discharge a battery in 1/n hours. For example, a rate of C /2 willdischarge a battery in 2 hours, and a rate of 5C will discharge a battery in 0.2 hours. Fora 2 Ah battery, the C /5 rate is 400 mA, while its 5C rate is 10 A.

As mentioned before, C depends on battery discharge current rate according to Peuk-ert’s equation. For a lead acid battery, the Peukert constant can range from 2.0 to 1.05depending on manufacturing technology.

Energy Stored (E)

The energy stored in a battery is dependent on battery voltage and the amount of chargestored within. The watt hour or Wh is the SI unit for energy stored. Assume a constantvoltage (CV) for the battery. Then

E (Wh) = V × C (11.1)

where V is the voltage and C is the capacity in Ah. The capacity of the battery changeswith the discharge rate, and the associated discharging current affects the voltage value.The energy stored is thus not a constant quantity and is a function of two variables,namely, the voltage and capacity of the battery.

State of Charge (SOC)

A key parameter in the electric vehicle is the SOC of the battery. The SOC is a measureof the residual capacity of a battery. To define it mathematically, consider a completelydischarged battery. The battery is charged with a charging current of Ib(t); thus from timet0 to t , a battery will hold an electric charge of

t∫t0

Ib (τ ) dτ

The total charge that the battery can hold is given by

Qo =t2∫

t0

Ib (τ ) dτ (11.2)


where t2 is the cutoff time when the battery no longer takes any further charge. Then,the SOC can be expressed as

SOC (t) =

t∫t0

Ib (τ ) dτ

Qo

× 100% (11.3)

Typically, the battery SOC is maintained between 20 and 95%.A common mistake that people may make about a battery’s charge is that when a battery

“goes dead,” the voltage goes from 12 to 0 V (for a 12 V battery). In reality a battery’svoltage varies between 12.6 V with a SOC of 100% to approximately 10.5 V with a SOCof near 0%. It is advised that the SOC should not fall below 40%, which correspondsto a voltage of 11.9 V. All batteries have a SOC vs. voltage curve which can be eitherlooked up from the manufacturer’s data or determined experimentally. An example ofan SOC vs. voltage curve of a lead acid battery is shown in Figure 11.2. Note that fora lithium-ion battery, the curve may be much flatter, especially for the mid-SOC rangeof 40–80%.

Depth of Discharge (DOD)

The depth of discharge (DOD) is the percentage of battery capacity to which the batteryis discharged. The DOD is given by

DOD (t) =Qo −

t∫t0

Ib (τ ) dτ

Qo

× 100% (11.4)

10 100400 503020 8060 709.0

90

13.0

11.5

10.0

12.0

10.5

Battery State of Charge (SOC) Percentile

9.5

11.0

12.5

Bat

tery

Vol

tage

(V

)

0% SOC 10.5 V

100% SOC 12.6 V

Figure 11.2 Example SOC vs. voltage curve for a 12 V battery


Generally, a battery is prevented from having a low DOD. The withdrawal of at least80% of battery capacity is regarded as a deep discharge.

One important precaution is that the charge in a battery should never be dischargeddown to zero voltage, otherwise the battery may be permanently damaged. So, in this case,a cutoff voltage is defined for the battery voltage so that the voltage at the battery terminalswill never drop below this cutoff voltage. This point is referenced as 100% DOD.

Specific Energy

Specific energy means how much electrical energy can be stored per unit mass of battery.The SI unit for this quantity is watt hour per kilogram. Knowing the energy stored andspecific energy of the battery, the mass of the battery can be easily obtained by dividingthe energy by specific energy. Again, the specific energy is not a constant parametersince the energy stored varies with discharge rate. A comparison of the specific energyof various energy sources (typical numbers) is given in Table 11.1.

Energy Density

Energy density means how much electrical energy can be stored per cubic meter of batteryvolume. It is computed by dividing the energy stored in the battery by the battery volume.The SI unit for energy density is watt hour per cubic meter.

Specific Power and Power Density

Specific power means how much power can be supplied per kilogram of battery. Note thatthis quantity is dependent on the load served by the battery and is thus highly variableand anomalous. The SI unit of specific power is watt per kilogram. Specific power is theability of the battery to supply energy. Higher specific power indicates that it can give andtake energy quickly. Volume specific power is also called power density or volume powerdensity, indicating the amount of power (time rate of energy transfer) per unit volume of

Table 11.1 Specific energy of different energy sources

Energy source Specific energy (Wh/kg)

Gasoline 12 500Natural gas 9350Methanol 6050Hydrogen 33 000Coal 8200Lead acid battery 35Nickel metal hydride battery 50Lithium-polymer battery 200Lithium-ion battery 120Flywheel (carbon fiber) 30Ultracapacitor 3.3


battery. If a battery has high specific energy but low specific power, this means that thebattery stores a lot of energy, but gives it out slowly. A Ragone plot is used to depict therelationship between specific power and specific energy of a certain battery.

Ampere-Hour (or Charge) Efficiency

Ampere-hour efficiency is the ratio between the electric charge given out during discharg-ing a battery and the electric charge needed for the battery to return to the previous chargelevel. In practice, these two values will never be equal, limiting the efficiency to 100%. Infact the typical values of charge efficiency range from 65 to 90%. The efficiency dependson various factors such as the battery type, temperature, and rate of charge.

Energy Efficiency

This important quantity indicates the energy conversion efficiency of the battery, whichdepends a great deal on the internal resistance of the battery. It can be computed as theratio of electrical energy supplied by a battery to the amount of charging energy requiredfor the battery to return to its previous SOC before discharging. The efficiency decreasesconsiderably if a battery is discharged and charged very quickly. Typically, the energyefficiency of a battery is in the range of 55–95%.

Number of Deep Cycles and Battery Life

EV/HEV batteries can undergo a few hundred deep cycles to as low as 80% DOD ofthe battery. Different battery types and design result in different numbers of deep cycles.Also, the usage pattern will affect the number of deep cycles a battery can sustain beforemalfunction. The United States Advanced Battery Consortium (USABC) has a mid-termtarget of 600 deep cycles for EV batteries. This specification is very important since itaffects battery life time in terms of deep-cycle number. So, generally, we should reducethe chances of DOD in the control strategy for EVs and HEVs in order to limit theoperating cost of the vehicles.

Example: The NiMH traction battery of the Toyota Prius 2004 model has thefollowing specifications:

• 168 cells (28 modules)

• 201.6 V nominal voltage

• 6.5 Ah nominal capacity

• 28 hp (21 kW) output power

• 1300 W/kg specific power

• 46 Wh/kg specific energy.


Assume that the voltage is relatively constant.

1. What is the energy rating of the battery in kilowatt hours?

Solution: 201.6 V × 6.5 Ah = 1310 Wh = 1.31 kWh

2. If the battery can be discharged at a maximum rate of 100 A, and only 40%can be discharged, for how many seconds can the battery be used when fullycharged?

Solution:(40%)6.5 Ah

100 A= 0.026 h = 93.6 seconds

3. If the battery can be charged at a maximum rate of 90 A, and the current SOCis 40%, how long does it take to charge the battery to 80% SOC?

Solution:(80% − 40%)6.5 Ah

90 A= 0.0289 hours = 1.73 minutes

4. The battery has an internal resistance of 0.15 �. What is the efficiency at maxi-mum charge rate?

Solution: η = 1 −(902

)(0.15)

(90) (201.6)= 93.6%

5. The battery has an internal resistance of 0.1 �. What is the efficiency at maximumdischarge rate?

Solution: η = 1 −(1002

)(0.1)

(100) (201.6)= 95%

6. How much voltage drop is caused by this internal resistance at maximumcharge/discharge?

Solution: At maximum charge, the voltage drop is 90 × 0.15 = 13.5 V; atmaximum discharge, the voltage drop is 100 × 0.1 = 10 V.

7. Does the efficiency change with maximum charge/discharge current?

Solution: Yes, the battery efficiency depends on the maximum charge/dischargecurrent and internal resistance.

8. If the leakage current is 20 mA, how many days does it take for the battery toself-discharge from 80% SOC to 40% SOC?

Solution:(80 − 40%)6.5 Ah

20 × 10−3 A= 130 hours = 5.4 days.

11.3 Comparison of Different Energy Storage Technologiesfor HEVs

Different energy storage technologies, including but not limited to those shown inTable 11.2, are available for HEV applications: Li-ion battery, nickel metal hydridebattery, lead acid battery, and ultracapacitors. In the table, the typical range of


Table 11.2 Comparison of energy storage technologies suitable for HEVs

Storage Cycle Efficiency (%) Specific Specifictechnology life power (W/kg) energy (Wh/kg)

Lead acid battery 500–800 50–92 150–400 30–40Li-ion battery 400–1200 80–90 300–1500 150–250Nickel metal hydride battery 500–1000 66 250–1000 30–80Ultracapacitor 1 000 000 90 1000–9000 0.5–30USABC long-term goals 1000 80 400 200

specification is given for each type of storage device along with the long-term goalset by the USABC (http://www.uscar.org/, http://en.wikipedia.org/wiki/United_States_Council_for_Automotive_Research). The advanced lead acid and Li-ion batteries are themost promising for application in HEVs. While battery and ultracapacitor technologieshave their respective advantages and disadvantages, hybridization could result in bettervehicle performance and longer battery life. The vehicle road load transients can behandled by ultracapacitors during acceleration and deceleration.

A battery is an electrochemical cell that can convert chemical energy into electricalenergy (redox reaction). There are three main parts in a battery: electrolyte, anode, andcathode. At the anode, the “negative” terminal, an oxidation reaction takes place and theelectrode loses electrons. At the cathode, the “positive” terminal, a reduction reactiontakes place and the electrode gains electrons. There is also a porous separator betweenthe two electrodes.

Lead Acid Battery

This type of battery is the earliest and the most widely used in automotive applications.For example, it is extensively used as the starting battery to provide “cranking amps” foran automobile’s starter motor. One of the plates is made of lead while the other plate ismade of lead dioxide. The electrolyte is composed of sulfuric acid. These batteries canlast a long time if they are charged and discharged properly.

The energy-to-volume ratio is low for a lead acid battery. This ratio can be usedto measure the drive range of an EV. How to treat used lead acid batteries is anotherserious problem. As a result, this type of battery cannot satisfy the requirements forfuture environmentally-friendly vehicles. Fortunately, lead acid batteries have very highrecyclability with good recycling infrastructure.

The chemical equation for a lead acid battery during discharge is

PbO2 + Pb + 2H2SO4 → 2PbSO4 + 2H2O (11.5)

The chemical equation for a lead acid battery during charge is

2PbSO4 + 2H2O → PbO2 + Pb + 2H2SO4 (11.6)

According to the chemical equation, electrolyte and the active material on the batteryplates are consumed and water and lead sulfate are produced when a lead acid battery is


discharged. On the other hand, during the charging process, electrical energy is absorbedby the battery, water and lead sulfate are consumed, and electrolyte and the active materialat the plates are produced.

Nickel Metal Hydride Battery

The NiMH battery is a new type of high-capacity battery. Its technology has grown rapidlyin the past five years. It has many advantages such as environmental friendliness, highspecific energy and energy density, and a long cycle life. The NiMH battery has alreadyoccupied a good market share as energy storage in HEVs.

The overall reversible chemical reaction occurring in a NiMH cell is:

MH + NiOOH ⇔ M + Ni(OH)2 (11.7)

Lithium-Ion Battery

In Li-ion batteries, Li ions alternatively move into and out of host lattices during charg-ing and discharging cycles. This fundamental mechanism has led to the Li-ion battery’snick-name of “rocking-chair” battery. In its physical composition, a Li-ion battery hasanode and cathode plates like a lead acid battery, except that these are made of lithiumcobalt oxide (or other lithium composites) and carbon. These plates and the separator areimmersed in a solvent which is most commonly ether [4]. This type of battery can bemade with very high energy density. The overall reversible chemical reaction occurringin a Li-ion cell is

LixC + Li1−xMyOz ⇔ C + LiMyOz (11.8)

Li-ion batteries do not have the “memory effect” that causes other rechargeable batteries tolose their maximum charge level when repeatedly recharged or not charged to full capacity.Li-ion batteries also impact the environment less due to their composition. Unlike leadacid batteries, they have a much lower self-discharge rate, thus greatly increasing idleperiod capabilities. These batteries also have a higher power-to-volume ratio which alsomakes them ideal for automotive applications [5]. Two of the latest EVs, the Nissan Leafand Chevy Volt, both use lithium batteries. Different materials can be used for anode. TheMn series Li-ion battery has been used in the Nissan Leaf, Mitsubishi i-MiEV, GM Volt,Chrysler S400 Hybrid, BMW 7 series ActiveHybrid, THINK TH!NK City, and HyundaiSonata Hybrid Blue Drive.

However, lead acid batteries have remained the favored ones due to cost and thefact that Li-ion batteries require a lot more safety attention. These batteries are muchmore susceptible to overcharging and overdischarging and the associated safety hazards.Overcharging or overdischarging the battery can severely damage the plates inside thecase. Overcharging can also cause gassing of the electrolyte and buildup of pressure in thecase, which can lead to an explosion, therefore a precise regulatory system is necessary.The reduction in life due to this effect is much greater than in lead acid batteries. Thesame reaction occurs when they are used improperly, leading to overheating and the riskof an explosion. In the event of charging or discharging a Li-ion battery, the voltagemust be monitored carefully because the absolute limits are so close to the required 100%SOC voltage [6].


Ultracapacitors

Ultracapacitors have a very long shelf life, with much lower maintenance requirements,enhanced performance at low temperature, and environmental friendliness. The onlydownside to ultracapacitors is their initial cost and relatively low energy density whencompared to batteries [7]. Unlike batteries, no chemical reactions are needed for stor-ing and retrieving electrical energy with ultracapacitors and thus the energy efficiency ishigher. The ultracapacitor’s SOC is easier to estimate than that of a battery because thevoltage is the only measurement needed (SOC is proportional to V 2). Also, ultracapacitorscan be charged to a specific value and, due to their shelf life and charging mechanism,they can hold that charge with virtually no loss. Batteries are incapable of achievingthis. Repeated depletion cycles of a lead acid or Li-ion battery can be detrimental to itslifespan; however, this is not the case with an ultracapacitor.

Ultracapacitors provide more freedom in the DC link voltage or wherever else theyare used because their charge does not depend on a certain voltage. Whatever voltagethey are charged to is what they retain [8]. As a result, a hybrid topology consisting ofultracapacitors is desired when variable voltages are required. This would be beneficialin portable fuel cell/ultracapacitor power supplies that could be used in emergencies orgeneral use. A vast array of loads or devices could be powered by this system.

Ultracapacitors allow rapid charging and discharging. This is especially useful forfaster and efficient regenerative energy recovery in HEVs as well as for rapid charg-ing of PHEVs. Simple charging methods can be used without needing a sophisticatedSOC detection algorithm, and there is little danger of overcharging so long as the volt-age is below the maximum allowable value. Ultracapacitors have a long cycle life (onthe order of a million cycles), with little degradation over hundreds of thousands ofdischarge/charge cycles. In comparison, rechargeable batteries last for only a few hundreddeep cycles.

The ultracapacitor’s energy density is much lower than that of an electrochemical bat-tery (3–5 Wh/kg for an ultracapacitor compared to 30–40 Wh/kg for a lead acid battery,and 120 Wh/kg or more for a Li-ion battery), and its volumetric energy density is onlyabout 1/1000th of that of gasoline. As in any capacitor, the energy stored is a functionof voltage squared. Effective storage and retrieval of energy requires complex electroniccontrol and balancing circuits involving power electronics switches. The self-dischargerate is much higher than that of an electrochemical battery and thus it is only suitablefor short-term energy storage. An enormous amount of energy could be released in afraction of a second from an ultracapacitor and this could be life threatening if pre-cautions were not taken. The internal resistance of ultracapacitors is very low, resultingin high cycle efficiency (95% or more). Environmentally, ultracapacitors are safer sincethey do not contain corrosive electrolytes or other highly toxic materials. In compari-son, the reactive chemical electrolytes of rechargeable batteries present a disposal andsafety hazard.

A comparison of the power density and energy density of different energy storagesystems (ESSs) is illustrated in Figure 11.3 (Ragone plot or chart).


10 102 104

Ene

rgy

Den

sity

(Wh/

kg)

Power Density (W/kg)103

10–1

1

10

102

103

ElectrolyticCapacitor

Ultra-Capacitor

Fuel cell Li-ion

Ni-MHLead Acid

10–2

Figure 11.3 Comparison of power density and energy density for ESS in HEVs

11.4 Modeling Based on Equivalent Electric Circuits

11.4.1 Battery Modeling

A commonly used simple battery model is shown in Figure 11.4. It consists of an idealbattery with open-circuit voltage Voc and a constant equivalent internal resistance Rint .The battery terminal voltage is Vt . Voc can be obtained from the open-circuit measurement,and Rint can be measured by connecting a load and measuring both the terminal voltageand current, at fully charged condition. The terminal voltage Vt can be written as

Vt = Voc − IbRint (11.9)

+

+−

−

Rint (SOC, T )

Ib

Voc Vt

Figure 11.4 Simple battery model


+

−

+−

Ib

Rd

Voc Vt

Rc

Figure 11.5 Battery model accounting for the different charging and discharging resistances

Voc, the equilibrium potential of the battery, is a nonlinear function of the SOC andtemperature T given by Larminie and Lowry [9]:

Voc = Eo + (RT/F ) × ln(SOC/(1 − SOC)) (11.10)

where E o is the standard potential of the battery, R the ideal gas constant, T the absolutetemperature, and F the Faraday constant

This is a fairly good way of predicting the battery voltage. However, the battery open-circuit voltage does not remain constant. As discussed above, the voltage is affected bythe SOC/DOD of the battery and temperature. The variation in the open-circuit voltagedue to DOD, for a lead acid battery cell, as given by Larminie and Lowry [9] is

Voc = (2.15 − DOD × (2.15 − 2.0)) (11.11)

The main drawback of this model is that it cannot capture the response to dynamic eventsin a battery. For example, if a load is connected to the battery, the terminal voltage willimmediately change to a new, lower value according to this simplified model. In fact thisis not true; rather it will take some time for the voltage to settle to a new value.

The internal resistance of the battery has different values on charging and discharging.To account for the different resistance values under charge and discharge conditions, thecircuit model in Figure 11.4 can be modified as shown in Figure 11.5. An improved modelcan be obtained by incorporating a self-discharge resistance Rp in parallel with Voc.

In this model, the internal resistance takes different values during the charging and dis-charging process: Rc for charging and Rd for discharging. The internal resistance is usedto model all energy losses within the battery during charging and discharging, includingelectrical and non-electrical losses. The ideal diodes are present only for modeling pur-poses of selecting either Rd or Rc as the internal resistance based on the current directionand have no physical significance in the battery. For a given required power, the batterycurrent I b is expressed as

Ib = Voc − √V 2

oc − 4 × R × Preq

2R(11.12)

where

R ={Rdischarge(SOC, T ) for Ib ≥ 0, Preq ≥ 0Rcharge(SOC, T ) for Ib < 0, Preq ≤ 0


+

+−

−

Ib

Rd

Voc Vt

Rc

Rb

C

Figure 11.6 Dynamic battery model

Vmin < Vt < VmaxCb = f (T )

Rc = f (T, SOC )

Re = f (T, SOC ) Rt = f (T, SOC )

Cc = f (T )

+

−

Figure 11.7 RC model of a battery

The sign of the battery current, I b , appears to be positive in discharging mode and negativein charging mode.

The models in Figures 11.4 and 11.5 have the disadvantage of not being sensitive todynamic events in the battery. In order to model such dynamic or transient effects in abattery, a capacitor is added to the model as a parallel branch, as shown in Figure 11.6.

The model in ADVISOR [3], given in Figure 11.7, is called the “RC model”; it takespower as an input and maintains the battery output voltage within the high- and low-voltage limits. The “RC model” can predict the average internal battery temperature as afunction of time while driving and during soak periods. In this model, the capacitor C b

is large enough to hold the capacity of the battery and the smaller capacitor C c is usedto reflect the dynamic changes in the battery.

11.4.2 Battery Modeling Example

Here, the model in Figure 11.5 is implemented in MATLAB as an example. The SOC isobtained by the following equation:

SOC = 1 − used Ah capacity

max Ah capacity(11.13)


where the used Ah capacity is calculated from

Used Ah capacity =

⎧⎪⎪⎨⎪⎪⎩

t∫0ib(t)dt for ib(t) ≥ 0 discharge

t∫0ηcoulombib(t)dt for ib(t) < 0 charge

(11.14)

where I b(t) is the charge/discharge current for the battery and ηcoulomb is the coulombicefficiency for charging.

In this example, Voc is a function of the SOC; the data is obtained from the experimentaltests of a Hawker Genesis 12 V, 26 Ah, 10EP sealed valve-regulated lead acid (VRLA)battery and implemented as a lookup table in MATLAB [3]. The open-circuit voltage ofeach cell takes values ranging from 11.7 to 12.89 V corresponding to the SOC from 0 to 1.The internal resistors Rd and Rc are also indexed by the SOC and T from two differentlookup tables corresponding to discharging and charging respectively. The coulombicefficiency is assumed to be 90%. Voc, Rd , and Rc of this battery model are plotted inFigure 11.8.

10

15

20

25

30

35

40

45

SOC

Rdi

scha

rge

(mill

iohm

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 111.5

12

12.5

13

SOC

VO

C (

Vol

t)

20

25

30

35

40

45

50

Rch

arge

(m

illio

hm)

SOC

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Figure 11.8 Voc, Rd , and Rc of the battery model


0 5 10 15 200.75

0.8

0.85

0.9

0.95

1

Time (s)

SOC

0 5 10 15 20–5

0

5

Time (s)

Bat

tery

Cur

rent

Ib

(A)

0 5 10 15 20200

202

204

206

208

210

Time (s)

Ter

min

al V

olta

ge V

t(V)

Figure 11.9 Simulation results of the battery model

A battery pack consisting of 16 battery cells in series is simulated to test the developedbattery model. The magnitude of battery current pulse is 5 A with a frequency of 0.5 Hzand pulse width of 1.6 seconds. The battery temperature is assumed constant at 34.7 ◦C.Simulation results are shown in Figure 11.9. From the discharge and charge current pulseprofiles, in the first 10 seconds discharge current pulses are applied to the battery model,while in the last 10 seconds charge current pulses of the same magnitude are applied.The battery pack terminal voltage decreases from around 206 V over the first dischargingduration in the first 10 seconds. In the last 10 seconds, however, the terminal voltage gainsalmost linearly over the charging intervals. From the SOC plot, it can be seen that in thefirst 10 seconds, the SOC is reduced from 1.0 to 0.78 during discharging but increases to0.975 after the pulsed charge currents are applied.

11.4.3 Modeling of Ultracapacitors

In any capacitor, the electrical energy is stored by using a positively charged electrodesurface and a negatively charged electrode surface with a dielectric separator betweenthem. In supercapacitors or ultracapacitors, special carbon-based electrodes are made to


provide an extremely large internal active surface area. The charge and energy stored ina capacitor are given by

Q = C × V (11.15)

E = 1

2CV 2 (11.16)

where

Q is the charge in the capacitor,C is the capacitance in farads,V is the voltage across the capacitor,E is the energy of the capacitor.

For ultracapacitors, the SOC can be computed with high accuracy:

SOC = C (V − Vmin)

C (Vmax − Vmin)= (V − Vmin)

(Vmax − Vmin)(11.17)

where Vmax and Vmin denote maximum and minimum allowable voltage for theultracapacitor.

The capacitance C of a capacitor is given by

C = εA

d(11.18)

where ε is the permittivity of the dielectric medium, A is the plate area, and d is thedistance between the plates.

Supercapacitors get their name from their ability to store high energy. This can be doneby increasing the area of parallel plates in a capacitor. The capacitance is increased bydecreasing the separation distance between the parallel plates. This is the key to moderncapacitors, but the voltage across the capacitor must be small otherwise the capacitormay be damaged and act like a short circuit. This puts a limit on the maximum energya capacitor can store. Connecting capacitors in series decreases the effective capacitanceand increases the voltage across them. The energy stored now increases, but not as thevoltage squared because of the problem of voltage imbalance among the cells. The fun-damental reason for the voltage mismatch between the series-connected cells is due tothe variation of capacitance between the cells. The problem can be avoided by charge-equalizing circuits. These circuits balance the charge on the adjacent capacitors keepingthe cell voltage the same.

The simplest equivalent electric circuit model for an ultracapacitor consists of a capac-itance in series with an internal resistance, which models all the energy losses in theultracapacitor. This model can capture to a large extent the essential behavior and dynam-ics of the ultracapacitor for most modeling and analysis needs. However, more detailedand complex models can be developed as outlined in the following.

A slightly more complex model is shown in Figure 11.10. The equivalent circuit of theultracapacitor unit consists of a capacitor (C), an ESR to model the internal resistanceduring discharging and charging, and an equivalent parallel resistance (EPR), Rp , which


v(t)Rp

ESR

i(t)

Figure 11.10 Simple electrical equivalent circuit of a supercapacitor

v(t)

i(t)

Figure 11.11 The fifth-order ultracapacitor model

is much higher than the ESR, to account for any self-discharging losses. These parameterscan be obtained from ultracapacitor data sheets provided by manufacturers.

For an accurate representation of an ultracapacitor, a fifth-order model is needed [10].The ultracapacitor is represented by a series–parallel connection of resistors and capac-itors, and the values of each can be varied to achieve the desired modeling accuracyover a wide range of operating frequency. Figure 11.11 shows the representation of anultracapacitor by a fifth-order model. For electric vehicle modeling and simulation, themodel order can be reduced to third order to account for a frequency range of interestbetween 0.0034 and 1.44 Hz [10].

11.4.4 Battery Modeling Example for Hybrid Batteryand Ultracapacitor

As mentioned previously, hybridization of battery and ultracapacitor could yield a bet-ter energy storage system to supply transient loads with the benefit of longer batterylife. Based on the battery model described in Section 11.4.2, an electrical model for ahybrid energy storage system (HESS) consisting of a battery and supercapacitor systems isdeveloped and implemented in MATLAB/Simulink to demonstrate the concept of passiveHESS. The equivalent circuit of the HESS is shown in Figure 11.12.

In Figure 11.12, the lead acid battery is modeled by the internal emf, E m , and theinternal resistor R2. The values of the internal series resistance R2 are dependent on thecharge/discharge current. R2,c and R2,d correspond to charging resistance and dischargingresistance respectively. In this HESS model, the internal parameters, including E m andR2, are SOC and temperature, T dependent.


Cc

RcR2,c

Em

i0

vR2,d

ibic

+

−+

−

Figure 11.12 Equivalent circuit model for HESS consisting of lead acid battery and ultracapacitor

Though the charge/discharge characteristic of the ultracapacitor can be modeled in morecomplex forms as shown in previous section, the ultracapacitor model used in this exampleis represented by a single lumped capacitance, C c , and a single lumped resistance, Rc . Inthis model, C c is the capacitance between the double layers of the ultracapacitor and Rc

models the internal loss inside the supercapacitor.For a given discharge current i 0, one can assume it is distributed between the battery

and ultracapacitor branches based on the impedances of the two branches, as denoted byi b and i c respectively in Figure 11.12. The calculation of i b and i c for a given i 0 will bedetailed in the following.

The Thevenin equivalent circuit of the HESS as shown in Figure 11.12 can be depictedas in Figure 11.13, where the Thevenin equivalent voltage and impedance of the HESS

+

−

IO (S )

Zth (S )

Vth (S )

Eth (S )

+−

Figure 11.13 Thevenin equivalent circuit of the HESS


circuit are denoted by E th (s) and Z th (s) respectively. Note that Figure 11.13 is derivedin the frequency domain and E th (s) and Z th (s) are obtained by

Eth(s) = Rc

R2 + Rc

Em

s + α

s(s + β)+ R2

R2 + Rc

Vc01

s + β(11.19)

Zth(s) = R2Rc

R2 + Rc

s + α

s + β(11.20)

where s is the complex frequency, Vc0 is the initial voltage across the ultracapacitor, and

α = 1

RcCc

(11.21)

β = 1

(R2 + Rc)Cc

(11.22)

By assuming the charge/discharge current is a pulsed signal with period T 1 and duty ratioD , the current for the first N pulses can be expressed as

i0(t) = I0

N−1∑k=0

[�(t − kT1) − �(t − (k + D)T1)] (11.23)

where I 0 is the magnitude of the current and �(t) is a unit step change function at t = 0.By applying the Laplace transform operation to this equation, one can readily obtain thecurrent in the frequency domain as

I0(s) = I0

N−1∑k=0

[e−kT1s

s− e−(k+D)T1s

s

](11.24)

Based on this, the internal voltage drop due to internal equivalent impedance inside theHESS, Vi(s), is given by

Vi(s) = I0(s) · Zth(s) = R2Rc

R2 + Rc

I0

[(α

β

1

s+ β − α

β

1

s + β

)× (

e−kT1s − e−(k+D)T1s)]

.

(11.25)

Thus, the terminal voltage of the HESS, Vth(s), can be calculated by

Vth(s) = Eth(s) − Vi(s) (11.26)

Performing the inverse Laplace transform, we get

v(t) = Eth(t) − vi(t) = Em + R2

R2 + Rc

(Vc0 − Em)e−βt

− R2I0

N−1∑k=0

[(1 − R2

R2 + Rc

e−β(t−kT1)

)�(t − kT1)

−(

1 − R2

R2 + Rc

e−β[t−(k+D)T1])

�[t − (k + D)T1]

](11.27)


Note that the second term in this equation is caused by the energy redistribution betweenthe battery and supercapacitor at the beginning of the discharge. This item decays as timelapses. The Thevenin voltage of the HESS finally approaches E m .

After obtaining the HESS terminal voltage, the charge/discharge currents flowingthrough the battery and the ultracapacitor can be calculated by the following twoequations respectively:

ib(t) = 1

R2[Em(t) − v(t)] (11.28)

ic(t) = i0(t) − ib(t) (11.29)

The above HESS model equations are implemented in MATLAB/Simulink. To verify thevalidity of this HESS model, a pulsed signal with a magnitude of 8 A, period T 1 = 2seconds, and duty ratio D = 0.8 is applied in the model. Note that positive values of i 0

correspond to discharge currents, while negative values correspond to charge currents. Thebattery pack consists of 16 Hawker Genesis 12 V, 26 Ah, 10EP sealed VRLA batteriesconnected in series to obtain 206.24 V at the original state. The ultracapacitor systemconsists of 77 modules of Nesscap 2.7 V/600 F capacitors connected in series to match theBESS terminal voltage. We assume that the ultracapacitors are fully charged at the initialstage having an initial voltage of 207.9 V. The internal resistance of the ultracapacitor is1m�. The simulation results are shown in Figure 11.14.

The discharge/charge current applied to the HESS is given in Figure 11.14a. Thebattery (BESS) current is shown in Figure 11.14b, from which one can find that theBESS current increases gradually to approach the pulse magnitude, 8 A, after five periodsor 10 seconds. The current balance between the applied current and that taken by theBESS is compensated by the parallel-connected supercapacitor, which can be found fromFigure 11.14c. This contribution of the supercapacitor will reduce the negative effect ofthe high discharge/charge current on the battery and hence effectively extends the lifetimeof the BESS. The terminal voltage and SOC of the battery over the simulated intervalare depicted in Figure 11.14d,e respectively. The exchange current between the BESSand supercapacitor is shown in Figure 11.14f. The exchange current originates from thesecond term in the equation for v (t), which is due to the difference between the initialvoltage of the ultracapacitor Vc0 and E m . The negative exchange current indicates thatthe ultracapacitor provides a decaying charging current to the battery. It can be observedthat the exchange current decays to 0 after about 25 seconds, which verifies the analysisregarding v (t).

11.5 Battery Charging Control

Charging technology plays a key role in maximizing battery performance. A proper batterycharging technique ensures battery safety and increases system reliability. The primaryrequirement of the charging process is to provide a fast and efficient way of chargingwithout degrading the battery. Some of the factors to be taken into account while chargingthe battery are:


0 10 20 30 400.65

0.7

0.75

0.8

0.85

0.9

0.95

1

Time (s)

Bat

tery

SO

C

0 10 20 30 40–6

–4

–2

0

2

4

6

Time (s)

Exc

hang

ed C

urre

nt (

A)

0 10 20 30 40

–2

–6

0

2

4

6

8

Time (s)

Bat

tery

Cur

rent

(A

)

–4

–80 10 20 30 40

Time (s)(a) (b)

(c) (d)

(e) (f)

HE

SS D

isch

arge

Cur

rent

(A

) 10

5

0

–5

–10

15

0 10 20 30 40

Time (s)

Supe

rCap

Cur

rent

(A

) 10

5

0

–5

–10

–150 10 20 30 40

200

202

204

206

208

210

Time (s)

Ter

min

al V

olta

ge (

V)

Figure 11.14 Simulation results of the HESS model: (a) applied discharge/charge current;(b) BESS current; (c) ultracapacitor current; (d) BESS terminal voltage; (e) SOC of the BESS;and (f) the exchange current between the BESS and the ultracapacitor

• Avoiding overcharging and undercharging.• Fast charging without affecting the battery life.• Maintaining a good quality of charging current.

Conventional charging methods include passive charging, constant current (CC) charg-ing, CV charging, and CC–CV charging. The pulse charging method is gaining popularity


because of its advantages over CC and CV charging. In this section, different chargingtechniques will be reviewed and discussed.

The simplest method is passive charging. This method is also the worst method forthe health of a battery. In passive charging, the battery is connected to a DC link whichhas a fixed voltage higher (hopefully by only a small percentage) than the rated volt-age of the battery. The charging current is unregulated in this method and can spikegreatly and possibly stay at a charging current above the safe value. As a result, there isthe risk of overcurrent and definitely overvoltage. This method is not recommended forcritical systems.

The second method is CV charging. For example, for a 12 V battery, the chargingvoltage would be maintained at a value of about 12.1 V regardless of the SOC. As aresult of CV charging, the current supplied to the battery is unregulated. In turn, witha low initial SOC there is a high current spike of unknown duration that could exceedthe safe charging current for the battery. This is a very fast method of charging with anexponential profile; however, the initial current spike can be damaging.

In the CC scheme, the charging voltage is varied and tracks the SOC of the battery tomaintain a CC at the battery’s DC link. As a result, the charging current profile of thebattery stays well below the safe charging rate. A great advantage of this method is thatit prevents heat from building up during the charging process and thus increases the lifeof the battery. The disadvantage of this method is that the constant higher current, evenat the end of the charging cycle, could lead to the growth of deposits which could shortthe plates in the batteries. As a result, the method is harmful at the end of the chargingcycle. This method also requires monitoring of both voltage and current measurements.It is slower than CV charging and has a linear charging profile.

In the CC–CV charging method, in the first stage a CC is maintained for charging thebattery. This current level is regulated at the safe charge limit by increasing or decreasingthe terminal voltage level of the charging source. Once the battery reaches a certain SOCor voltage value, the current is decreased (voltage decreased) to prevent damaging currentsat the end stage. This particular charging stage is usually called the floating charge. Thismethod combines the best attributes of both the CV and CC methods and eliminates theirdisadvantages. It is slower than the previously mentioned methods; however, it is thesafest for the battery [11].

The final charging method to mention is the pulsed current method. This method issimilar to pulse width modulation in that the pulse duration is proportional to the SOC.When the SOC is low, the pulses have a longer duration; when the SOC approaches100%, the pulse duration is reduced to near zero. At any pulse duration, there willbe a rest period in which the charging current is low. The relaxation period betweenpulses equalizes the chemical reaction in the battery as the ions diffuse and distributeevenly throughout the battery. This process normalizes the ion concentration in thebattery and thus prevents the negative effect experienced in CC charging. Because ofthe equal charge distribution, battery performance as well as battery life are enhanced.The charging rate can be controlled by varying the width of the pulses. This type ofcharging is advantageous over CC charging in that the charging rate is higher and batteryinternal impedance is kept lower [12]. This is the fastest method of the above-mentioned


10 100400 503020 8060 709.0

90

13.0

11.5

10.0

12.0

10.5

Battery State of Charge (SOC) Percentile

9.5

11.0

12.5

Bat

tery

Vol

tage

(V

)

0.0

Max Rated Current

Charging C

urrent (A)

0% SOC = 10.5 V

100% SOC = 12.6 V

80% SOC = 12.4 V

Figure 11.15 Plot of the battery voltage and charging current vs. SOC for a three-stage charger

methods and is becoming more popular with charging protocols. However, it wouldrequire a power electronics converter with a dedicated control algorithm.

The three-stage charger is an example of a controlled charging method on the markettoday and is mainly used for 12 V lead acid batteries. The CC–CV scheme is used. Thefirst stage is called the bulk charge stage. In this stage, the maximum current that thebatteries are rated for is delivered until the SOC reaches about 80–90%. The chargingvoltage at this stage exists between 10.5 and 15 V. The second stage, called the absorptionstage, delivers reduced current and maximum voltage in the range of 14.2–15.5 V. Thethird stage, the float charge stage, occurs when the batteries have reached nearly 100%SOC and all that is needed is to keep them at full charge. In this stage the voltage canvary between 12.8 and 13.2 V [13]. A sketch of a three-stage charger for a 12 V lead acidbattery is shown in Figure 11.15. An example of battery charge and discharge versus theSOC of a Li-ion battery is shown in Figure 11.16.

11.6 Charge Management of Storage Devices

The voltage of energy storage devices in electric vehicles is designed to be in the rangeof 300–400 V in order to provide adequate power for traction motors. As a result, manycells must be connected in series. When dealing with long battery or ultracapacitor strings,charge distribution must be equalized between the individual cells in order to preventunder- or overcharging and extend energy storage life. For HEV batteries, the cells willundergo many charge/discharge cycles, where major amounts of power are consumedfrom and injected into them [14]. In HEVs, the battery bank is charged with powerfrom the IC engine or via regenerative braking. The charge/discharge lapses can cause


(a)

(b)

0 0.2 0.4 0.6 0.8 110

10.5

11

11.5

12

12.5

13

13.5

DOD

Dis

char

ge v

olta

ge (

v)

C/20

C/8

C/4

C/2

3C/4

C

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 17

8

9

10

11

12

13

14

SOC

Vol

tage

('V

')

Figure 11.16 Lithium-ion battery discharge and charge characteristics. (a) Voltage versus depthof discharge (DOD) during the discharge of lithium ion battery at different discharge rates, at roomtemperature. (b) Battery voltage versus state of charge of a lithium ion battery, calibrated throughtesting data. (c) Charge current during pulsed charge of lithium ion battery. (d) Terminal voltageduring pulsed charge using the charging current shown in (c)


0 0.5 1 1.5

(c)

(d)

2 2.5 3 3.5x 104

0

5

10

15

20

25

Time ('s')

0 0.5 1 1.5 2 2.5 3 3.5x 104

Time ('s')

Cur

rent

('A

')

6

4

2

8

10

12

14

16

Bat

tery

Vol

tage

('V

')

Figure 11.16 (continued )

lasting damage to an unprotected battery string. A battery cell usually can act differentlyfrom other cells in a string. Some may charge or discharge faster than others due todifferences in internal impedance, temperature, and so on. This can cause an imbalance incharges among individual cells. If left unchecked, some can become over/underchargedor over/underdischarged. As a result, battery longevity is severely shortened and theovercharged batteries in an unprotected string could overheat, build up gas pressures, andexplode. The battery string presents a safety hazard and must be carefully maintained[6, 15–17]. Thus, a carefully controlled charge equalization system is needed. Our


discussion is focused on the battery system; a good discussion on cell balancing for anultracapacitor system can be found in [18].

The general principle behind the charge equalization circuit is to distribute an equalamount of energy or voltage to every cell in the series-connected string. In order to do this,the circuit must remove the energy from the more strongly charged cells and redistribute itto the weaker cells in order to obtain a uniform voltage or uniform energy distribution. Thisprocess is not instantaneous, it will happen over time, the length of which is determined bythe method used. Of course, this can be done in many different ways. Charge equalizationmethods fall into several different categories. The two top categories are dissipative andnon-dissipative equalizers. In the dissipative charge equalizing method, resistive shuntsare used. An active version of the resistive shunt method utilizes a PWM control switchin line with the resistive shunt. This method is the simplest but very inefficient.

Moving on to non-dissipative equalizers, there exist three more tiers in classification.First of all, there is the charge type, discharge type, and the charge–discharge (bidirec-tional) type. With respect to the charge–discharge type, the classification splits againinto current-fed equalizers and voltage-fed equalizers. For the former case, the equaliz-ers contain bidirectional converters; and for the latter case, the circuit contains switchedcapacitors. An equalization scheme using switched capacitors is shown in Figure 11.17.

The circuit in the figure is the double-tiered method. In this method, as in the single-tiered method, no dedicated control system is necessary. The switches are set to a fixedfrequency and are all switched at the same time. As a result, only one switching signalis necessary for this configuration. Even as the batteries reach an equalized state, theswitching action remains; however, this consumes negligible energy. The addition ofthe second tier allows batteries 1 and 3 to also exchange charge between each other.This allows for faster equalization. Another advantage is that sensors are not necessary,resulting in cost savings [19].

As for the current-fed bidirectional converter type, the system presented by Leeand Cheng [5] utilizes two bidirectional converters sharing one capacitor shown inFigure 11.18.

C1

C2

C3B2

B3

B1

Control

+

−

+

−

+

−

Figure 11.17 Circuit diagram of non-dissipative equalization with switched capacitors


VB1 VB2

L1L2

Q1

D1Q2

D2

+

+

+−

−+

−

+−

−C1

Figure 11.18 Circuit diagram of the current-fed bidirectional converter method

In the process of equalization, if battery 1 has a higher voltage than battery 2, thenthe first switch turns on at the duty cycle specified by the control system depending onthe system’s current state and measured values. This action charges the first inductorand then releases that energy through the capacitor and into the second battery, chargingit. Hence the method is named a current-fed charge–discharge type. When the secondbattery has a greater voltage than the first battery, the same process happens, just withreversed currents, with the first battery being charged. PWM signals for controlling theconverters’ switches regulate the amount of current going into and out of each batterycell. This would be dependent on the SOC of the batteries. Different methods can bedesigned to control the PWM signals.

With respect to the discharge type, there exist two more tiers, namely, direct transferand indirect transfer. With direct transfer, a serial recovery method uses step-up convertersand a parallel recovery method uses primary multiple windings. With indirect transfer, atwo-step method uses buck–boost converters and a multi-step method uses unidirectionalconverters. As with the charge type, there also exist two more tiers. The first is anautomatic method which uses secondary multiple windings. The second is a selectivemethod which uses secondary multiple windings plus switches [14].

11.7 Flywheel Energy Storage System

Flywheels are becoming of increasing interest in hybrid vehicle design, particularly forlarger passenger transit vehicles. This is because of the following reasons. First, therequirements on specific power and specific energy of the battery can be decoupled,affording optimization of the battery’s specific energy density and hence cycle life. Second,as the high-rate power demand and high-current discharge are greatly reduced by the loadleveling effect of the flywheel, the usable energy, endurance, and battery cycle life can beincreased. Third, the flywheel can allow rapid interim recharges with high efficiency duringperiods of low power demand or regenerative braking. Due to the combined effect of loadleveling of the main energy source and improved energy recovery during regenerativebraking, the range of the vehicle can be remarkably extended. The energy density isprimarily related to the flywheel’s speed of rotation. Increasing the speed of rotationproduces improved specific energy, but increases the potential safety hazard, and alsothe cost, since special bearings and high-strength materials are required. Instead of usinga battery or fuel cell, an EV can potentially be powered solely by an ultrahigh-speedflywheel. The corresponding long-term potential benefits for EV applications are possible,


since it can potentially provide higher specific energy and higher specific power than anybatteries. Its specific power may possibly be even higher than the IC engine. The flywheelshould also mitigate the problem of limited cycle life suffered by other sources becausethe life cycle of a flywheel is practically unlimited or at least longer than vehicle life[20]. The mechanical energy stored in a flywheel can be expressed as

E = 1

2Jω2 (11.30)

where J is the moment of inertia and ω the rotational speed. The above equation clearlyindicates that there is a squared relation between energy stored in the flywheel and itsrotational speed. Several companies have developed practical flywheel storage systems,like CCM of the Netherlands and Magnet Motor in southern Germany, and have accu-mulated considerable experience in the running of production flywheel–electric hybridvehicles [21]. CCM’s flywheels rotate at speeds of 15 000 rpm and operate with regener-ative electronic drives of modern design; energy storage efficiencies of up to 93% havebeen achieved [21].

Some basic concepts related to flywheels are important to consider. As noted above,a flywheel involves the need to transfer one form of energy into another, that is, thekinetic energy of the flywheel. In some cases electrical energy may be converted andstored in a flywheel in mechanical form. But if mechanical energy from one body isneeded to be transferred into another, that is, to a flywheel in this case, then it will benecessary to convert motion of the first body into motion of the flywheel. Consider thecase of regenerative braking in a hybrid vehicle. Here it is necessary for motion of thevehicle’s wheel to be reduced and the energy of motion to be transferred to the flywheel.In principle it is possible to transfer motion from one wheel to another as indicated inFigure 11.19.

In this figure, J v and ωv represent the inertia and angular velocity of the vehicle’swheel, and J FW and ωFW represent the same quantities for the flywheel. If the energyfrom the wheel is to be transferred to the flywheel, and assuming there is no loss in theprocess, at the end of the energy transfer the following relationship should hold:

1/2Jvω2v = 1/2JFWω2

FW (11.31)

Vehicle wheel

Jv, wv JFW, wFW

Flywheel

Figure 11.19 A simple connection between the vehicle’s wheel and the flywheel


cd

Jv, wv

D

JFW, wFW

Flywheel

AB

C

a

b

Vehicle wheel

Figure 11.20 Use of gears to mechanically transfer motion between the vehicle’s wheel and theflywheel

where ωv is the original angular velocity of the wheel and ωFW is the final angularvelocity of the flywheel after the energy transfer is complete. When the flywheel speedhas satisfied the above equation, the full vehicle wheel energy will be transferred to theflywheel, and the wheel’s speed should then be zero, theoretically speaking.

The question is: how is the motion transferred? If the transfer is to be done purelymechanically, then one can introduce two sets of gears in the vehicle wheel shaft andthe flywheel shaft respectively and slowly transfer the motion in a seamless manner asshown in Figure 11.20.

The idea is that initially the gears Aa will mesh together and other gears will bedisplaced laterally by some gear shifting scheme so that they are not engaged. With gearA having a very small diameter and gear a having a very large diameter, the flywheelwill slowly start moving. The angular velocity will be such that the energy is conserved,that is, if the initial velocity of the wheel was ωv , then after Aa meshing, the relationshipat steady state should be

1/2Jvω2v = 1/2JFW ω2

FW + 1/2Jvω2v(new) (11.32)

The value of ωv(new), the vehicle wheel velocity after some transfer of motion, will obvi-ously be smaller than the original value ωv . Next time the gear Aa can be disengaged (bysliding the gears sideways) and Bb can be engaged. Following the same logic, the speedof the flywheel now increases a little more and the speed of the wheel will be reduced alittle more. Similarly, continuing with the process of engaging Cc, and then Dd , and eachtime disengaging the previous gear set, the speed of the wheel will reduce more and moreand the speed of the flywheel will increase ultimately until the energy of the flywheelwill be equal to the energy of the wheel at its original speed prior to motion transfer. Inprinciple this will be best achieved if the gears are continuously variable, rather than adiscrete set of gears as in Figure 11.20. Obviously, due to the fact that an infinite numberof gears are not possible, the process will be achieved through some discrete jumps ifFigure 11.20 is followed. Also, note that the final energy of the flywheel will be somewhatless than the original energy of the wheel due to losses in the system.


FlywheelVehicle wheel

GeneratorMotor

Converter /Inverter

Generator Motor

Motion transferdirection

Jv, wv JFW, wFW

Figure 11.21 Electromechanical means to transfer motion between the vehicle’s wheel and theflywheel

The other way of transferring motion is by electromechanical means but using thesame principle as above. Here a motor generator scheme is used, which provides thecontinuously variable speed ratio and motion transfer through electromechanical means.This is illustrated in Figure 11.21.

In this figure, if the motion is to be transferred from the wheel on the left to theflywheel on the right, then the left electric machine will be a generator, and the powerconverter will convert the generator power properly to energize the other electric machineon the right. This one will then become a motor and will accelerate the flywheel. Themotor generator sets and the power converter serve as a continually variable transmissionif one wants to see it from that perspective. Eventually the speed of the wheel will bereduced and the speed of the flywheel will increase by an amount such that the reductionof mechanical energy of the wheel will be equal to the mechanical energy increase of theflywheel, less any losses. If it is intended to accelerate the vehicle again, the process willbe reversed.

One final note regarding the flywheel is that, when it is rotating, it can cause a gyro-scopic effect if the vehicle suddenly tries to turn quickly. This problem can be generallyremedied by having two rotational masses with the same inertia but moving in oppositedirections. This can be easily implemented by placing the two rotating members in asingle housing with a gear in between.

11.8 Hydraulic Energy Storage System

Energy can also be stored by using a hydraulic system, where it is stored in the form ofa compressed fluid or gas in a cylinder or similar means, known as an accumulator. Topressurize a compressible fluid, one needs mechanical power and energy, which can comefrom an IC engine or any other engine used to activate a hydraulic pump. Of course, inthis case the original source of energy which drives the engine is chemical energy ofthe gasoline or diesel. While extracting the energy back from the hydraulic storage, onecan use a hydraulic motor. The system-level scheme for realizing a hydraulic energy


Gasoline,diesel etc.

IC Engine Hydraulicpump

Reservoir with fluid athigh pressure

Reservoir with fluidat low pressure

Hydraulicmotor

Mechanicalload

Compressible fluid forenergy storage

Figure 11.22 Generic scheme for hydraulic energy storage and extraction

storage and extraction mechanism is shown in Figure 11.22. In this figure, the dashedlines indicate the fluid flow path.

In Figure 11.22, chemical energy of the gasoline or diesel will drive an IC engine. Theengine will drive a hydraulic pump which will basically draw an incompressible fluidfrom a low-pressure reservoir and increase its pressure. The high-pressure incompressiblefluid can be used to drive a hydraulic motor which can drive some mechanical load. In thehydraulic motor, the mechanical fluid enters at high pressure and exits at low pressure,doing mechanical work in the process. Upon exiting the hydraulic motor, the low-pressurefluid flows to the low-pressure reservoir and the fluid flow circuit is completed. The high-pressure reservoir fluid can also move a piston or similar mechanism which in turncan push against a compressible gas. This compression will cause energy to be storedin the gas. Once some predefined pressure has been achieved, a valve can be used toprevent further pressurization of the gas. While extracting energy from the accumulator,an appropriate valve can be opened and the compressed gas will work against some pistonand pressurize the incompressible fluid in the high-pressure reservoir, which in turn willdrive a hydraulic motor.

In Figure 11.22, it can be seen that the dashed fluid flow path forms a closed loopsystem. It can also be seen that the fluid path can be bidirectional between the accumulatorand the high-pressure fluid reservoir. Note that energy itself is stored in the compressiblegas or fluid. The incompressible fluid (liquid) provides a flexible path or actuator, whichreplaces any mechanical linkage which otherwise would have been used for actuation.

Note also that an accumulator by itself is not of much use without the peripheralequipment or subsystems shown in Figure 11.22, since ultimately the goal is to be ableto use the stored energy in a beneficial manner when needed, or, when extra energyis available, to store it in the accumulator. All the above items shown in Figure 11.22together form the overall hydraulic energy extraction and storage system mechanism.Further details on hydraulic systems are given in Chapter 6.

11.9 Fuel Cells and Hybrid Fuel Cell Energy Storage System

11.9.1 Introduction to Fuel Cells

Research and development have been carried out on fuel cells as possible alternative powersources for a wide range of applications, from portable power sources, to the distributed


High-Output Battery

Stores hydrogen gascompressed at extremelyhigh pressure to increasedriving range

Stores energy generatedfrom regenerative brakingand provides supplementalpower to the electric motor

Vehicles image courtesy ofAmerican Honda Motor Co., Inc.

Converts hydrogen gasand oxygen intoelectricity to power theelectricmotor

Governs the flow of electricity

Power Control Unit

Electric MotorPropels the vehicle much morequietly, smoothly, andefficiently than an internalcombustion engine andrequires less maintenance

Fuel Cell Stack

Hydrogen Storage Tank

Figure 11.23 Honda 2008 FCX Clarity fuel cell car. (Source: http://www.fueleconomy.gov/feg/fuelcell.shtm. Courtesy American Honda.)

generation (DG) of electricity, to use in a new generation of fuel cell-powered electricvehicles, and so on. There are different types of fuel cells, such as the proton exchangemembrane fuel cell (PEMFC), alkaline fuel cell, phosphoric acid fuel cell, molten carbon-ate fuel cell, solid oxide fuel cell, and direct methanol fuel cell. Among them, PEMFCsare primary candidates as power sources to drive next-generation vehicles, mainly due totheir relatively low operating temperature (around 80 ◦C). In the PEMFC, hydrogen fuelis converted to electricity based on an electrochemical process involving the use of pre-cious catalytic materials such as platinum. Some fuel cell prototype vehicles have alreadybeen built for the purpose of proof-of-concept and commercial demonstration. Examplesinclude the GM HydroGen3, Honda 2008 FCX Clarity (as shown in Figure 11.23), ToyotaFCHV-adv, among many others. Many fuel cell buses have been developed and are inoperation around the world for demonstration purposes; an example fuel cell bus is shownin Figure 11.24.

High cost, unsatisfactory durability, poor transient performance, and subzero temper-ature startup issue are the main obstacles for the commercialization of clean fuel cellvehicles. Further, current fuel cell systems do not allow bidirectional energy flow andthus have difficulty in recovering braking energy. Therefore, some kind of hybridizationof fuel cells with other energy storage devices such as batteries and ultracapacitors willbe advantageous for a long period of time. For example, the Toyota FCHV fuel cellvehicle uses a NiMH battery pack as the secondary energy source, and the Honda FCXfuel cell vehicle uses ultracapacitors as an energy buffer to achieve powerful, responsivedriving [22, 23]. In hybrid powertrains, the fuel cell system provides the base power forconstant speed driving, while other energy storage devices provide additional peak powerduring acceleration and high-load operation and recover braking energy by regeneration.


Courtesy: ISE Corporation Photo Gallery, Cuong Huynhhttp://www.isecorp.com/gallery/albums.php

Figure 11.24 The AC Transit Fuel Cell Bus – San Diego, 2005. (Courtesy ISE Corporation.)

Hence, the fuel cell system power rating and cost will be reduced; the powertrain transientperformance will be improved; and energy efficiency will be increased.

There have been some studies and experiments involving hybrid fuel cell vehicles[22, 24, 25]. In [22], the requirements on energy storage devices in a fuel cell vehicleare analyzed. A mid-size SUV and a mid-size car are designed and simulated usingADVISOR in order to help the FreedomCar technical teams properly size the energystorage devices onboard fuel cell vehicles. It is concluded that the powertrain cost andvolume can be greatly reduced over the pure fuel cell vehicle due to fuel cell downsizingvia hybridization. In [24], a fuel cell system with a nominal power of 48 kW is hybridizedwith supercapacitors with a storage capacity of 360 Wh. The hybrid system is implementedon a road vehicle and tested. The test results demonstrate that good transient performanceand impressive energy efficiency are achieved. In [25], optimization tools are linked toADVISOR for the optimal design of a battery fuel cell SUV. In particular, ratings of fuelcell and battery and energy management strategy are optimized to maximize fuel economy,while meeting the pre-specified vehicle performance constraints. For a US city/highwaycomposite test procedure, the optimizer chose a 66 kW fuel cell system and a 28-modulebattery pack with a large ampere-hour capacity of 50 Ah per module.

A fuel cell may be considered as an “open” battery in that the energy capacity is notlimited by the reductant and oxidant contained within a cell. Instead, the energy generatingcapacity is determined by the amount of onboard hydrogen fuel. As a power source, afuel cell can be three times more efficient (typical efficiency value of 60%) than an IC


engine (typical efficiency value of 20%) [23] because the fuel cell is not subject to the“Carnot cycle” efficiency limit [26, 27]. Except for water as the only by-product, thereare no CO2 and other harmful emissions. Thus, a fuel cell is a clean energy source.However, current fuel cell systems still have a low power density (10–100 times lower)compared to combustion engines. The short-term goal of power density for fuel cellsystems is 0.5 kW/l [22], while current IC engines can reach up to 50 kW/l. The fuelcell vehicle drive range is still much shorter than conventional vehicles and is limited bythe hydrogen storage energy density and cost. In addition, the fuel cell system has thedisadvantages of slow startup and slow power response. Hence, a pure fuel cell vehiclehas an unsatisfactory acceleration performance. Further, a fuel cell system cannot makeuse of any braking energy for improving fuel economy and driving range because of itsinability to regenerate energy. Another characteristic of the fuel cell system is that itsefficiency peaks near 25% of the rated power, with relatively lower efficiency at low andhigh output power. Thus, a vehicle control strategy should avoid the fuel cell system’slow-efficiency operating regions.

The general construction of a fuel cell consists of two plates between which a membranelayer and catalyst are compressed. One plate serves as the anode and the other plate is thecathode. These plates have channels which are etched in the material and let the reactantfuels uniformly make contact with the catalyst. In the case of the PEMFC, these platesare made of carbon and the polymeric ion exchange membrane is coated with a platinumcatalyst. The interconnections of the PEMFC are also made of carbon or some other typeof metal. This cell can generate approximately 0.6 V DC. In order to increase the currentcapabilities, the surface area of the plates is increased. This also results in a power increase.

Similar to batteries and ultracapacitors, many individual fuel cells are connected inseries/parallel to form a fuel cell stack to achieve the desired voltage, current, and powerratings for a particular application. A fuel cell system is composed of more than just thestack and the hydrogen and oxygen fuel. A complete fuel cell system is composed ofblowers, pumps, fans, ejectors, turbines, compressors, valves, and regulators. All thesecomponents work together to deliver and regulate the proper amounts of fuel and ensuresystem safety. This management system is known as the balance of plant or (BOP). Themain objectives of the BOP are hydrogen fuel preparation, air supply, thermal manage-ment, and water management. Fuel cell performance depends on the following parameters:temperature, reactant gas stoichiometric flow rates, anode and cathode pressures, anodeand cathode humidification.

Once the reactant fuels flow through the channels and come into contact with thecatalyst, a chemical process takes place. On the anode side (+), pressurized hydrogenflows through the network of small pathways. Through these pathways the hydrogencomes into contact with the platinum catalyst and splits into positive hydrogen ions andnegative electrons. The polymer electrolytic membrane only allows positive ions to flowthrough. As a result, the hydrogen ions pass through the membrane and the electrons areforced to pass through an external electric circuit. On the cathode side (−), the oxygengas (or ambient air) is passed through its own set of pathways. When it comes intocontact with the catalyst, the bonded oxygen atoms split into two independent atoms.Once the hydrogen ions have passed through the membrane, they join up with the oxygenatoms. Two hydrogen ions bond with one oxygen atom and two electrons pass throughthe cathode. As a result, the “exhaust” of the PEMFC is water vapor and heat [26, 27].


Figure 11.25 A 300 W fuel cell stack with BOP controller and hydrogen connections

Figure 11.25 shows a fuel cell power source with a power rating of 300 W, made byHorizon Fuel Cell Technologies. It has a maximum current output of 7 A at 42 V.

The fuel cell system contains the controller, electrical connections, rubber hydrogenlines for secondary hydrogen connections, primary valve, and purge valve. Details con-cerning the physical characteristics of the stack are given in Table 11.3. A 12 V source isrequired to power its BOP controller.

11.9.2 Fuel Cell Modeling∗

The relationship between cell current and voltage is modeled as the polarization curveand is based on the fuel cell’s steady state performance. In particular, the fuel cell outputvoltage is modeled as the thermodynamic potential with three types of overvoltages sub-tracted: activation overvoltages, ohmic overvoltages, and concentration overvoltages. Theactivation-related losses, or kinetic regime, are caused by the chemical reactions them-selves. The chemical reactions are not instantaneous when the reactants come into contactwith the catalyst; that is, the reaction must first be activated. This is mainly due to thesurface area of the catalyst. Ohmic losses are due to electrical resistances in the leads,connections, and the plates of the stack. This type of loss is greatly affected by heat, thesurface area of the plates in the stack, and the types of materials used in the fuel cellsystem. As for the concentration overvoltage or transport losses, these are caused by thelimit on mass transport rate of hydrogen and oxygen reactants.

In practical application such as powertrains of land-based vehicles, the output powerfrom the fuel cell system undergoes large variations, especially during acceleration and

∗ © [2008] Inderscience. Reprinted, with permission, from the International Journal of Electric and Hybrid Vehicles.


Table 11.3 List of technical specifications for the Horizon Fuel Cell

Technical specifications

Type of fuel cell PEMNumber of cells 72Rated power 300 WPerformance 43 V at 7 AH2 supply valve voltage 12 VPurging valve voltage 12 VBlower voltage 12 VReactants Hydrogen and airExternal temperature 5–35 ◦CMaximum stack temperature 65 ◦CComposition 99.999% dry H2

Hydrogen pressure 5.8–6.5 psiHumidification Self-humidifiedCooling Air (integrated cooling fan)Weight 2 kgDimensions 32.4 cm × 10.9 cm × 9.4 cmHydrogen flow rate 3.9 l/m at max powerStack efficiency 40% at 43 VController weight 250 gLow-voltage shutdown 36 VOvercurrent shutdown 12 AOver-temperature shutdown 65 ◦C

deceleration. During such transient operating periods, due to the existence of the double-layer capacitance at the interface between the electrodes and the electrolyte, the reactantgas manifold filling dynamics, cell surface dynamics, and other effects, the fuel cell stacksystem’s performance cannot be adequately represented by a pure algebraic steady statemodel. Hence, a dynamic fuel cell model is needed to provide more accurate predictionsof fuel cell system performance for the dynamic simulation and analysis of the fuel cellpower system. Such a model is illustrated in Figure 11.26.

In this equivalent circuit model, E cell represents the internal potential of the fuel cell.The output of fuel cell voltage Vd is E cell with three types of voltage drop subtracted:activation voltage drop, ohmic voltage drop, and concentration voltage drop. Activationvoltage drop can be separated into two parts: the one affected by internal temperatureand the other one caused by the equivalent resistance of activation. Capacitor C is theequivalent capacitor due to the double-layer charging effect. The formula for calculatingthese parts is as follows [28, 29]:

Vd = Ecell − Vact1 − Vact2 − Vconc − RohmicI (11.33)

whereEcell = E0,cell + RT

2Fln

[P ∗

H2

(P ∗

O2

)0.5]

− kE (T − 298)

and where E 0,cell is the standard reference potential at standard state (298 K and 1 atmpressure), R is the gas constant, and F is the Faraday constant. PH2

and PO2represent


Rohmic

Vd

I

Vc

Ract2

Rconc

Ecell − Vact1

C

−

+

−

+

+

−

Figure 11.26 A dynamic fuel cell model

the partial pressure of H2 and O2 activated inside the fuel cell. The asterisk * representsthe effective value of the parameter. k E is an empirical constant in volts per kelvin andT is the actual temperature when the fuel cell is operating. Thus

Vact1 = η0 + a (T − 298) (11.34)

where η0 is the temperature-invariant part of Vact1 and a is an empirical constant in voltsper kelvin. Similarly,

Vact2 = Ract2I (11.35)

Vconc = RconcI (11.36)

VC =(

I − CdVC

dt

)(Ract + Rconc) (11.37)

At steady state, the static characteristics of the fuel cell are as depicted in the followingequation, in which the relationship between the output voltage and current of the fuelcell can be approximately considered as linear by ignoring Vact2 and Vconc due to smallvalues of the two parameters at steady state:

Vd = Ecell − Vact1 − RohmI (11.38)

At steady state, E cell , Vact1, and Rohm can be considered as constants; therefore, the aboveequation can be modified approximately with the following expression:

Vd = K1 + K2I (11.39)

whereK1 = Ecell − Vact1 and K2 = −Rohm

Typical values of the parameters of the fuel cell model are given in Table 11.4.Figure 11.27 shows I –V polarization characteristics and the power–current curve for

a typical PEMFC model.


Table 11.4 Parameters of a typical fuel cell

E 0,cell (V) R(J/(mol K)) F (C/mol) PH2(Pa) PO2

(Pa)

58.9 8.3143 96 487 1.5 1.0k E (V/K) η0(V) a(V/K) Rohm (�) T (K)0.00085 20.145 –0.1373 0.2793 307.7

0 5 10 15 20 25

100

0

200

300

400

500

600

Load Current (A)

PEM

FC O

utpu

t Po

wer

(W

)

5 10 15 20 2522

24

26

28

30

32

34

36

38

40

Load Current (A)

Out

put

Vol

tage

(V

)

Figure 11.27 Fuel cell characteristic curve

11.9.3 Hybrid Fuel Cell Energy Storage Systems

The continuous power output of PEMFCs is satisfactory, but the voltage regulation ispoor and their response to instantaneous transient power load requirements is often toosluggish. To meet such peak power requirements economically, that is, to eliminate fuelcell overdesign, the PEMFCs are hybridized with batteries or ultracapacitors. Fuel cellshave a high specific energy while supercapacitors have high specific power, with secondarybatteries lying in between. The fuel cell/ultracapacitor hybrid system will ideally provideboth high power and energy densities at low cost. Further, since supercapacitors havelimited energy storage capability compared to batteries, a fuel cell/battery/supercapacitorhybrid system will have better power supplying capability. In vehicular applications, thebattery is also needed to start up the fuel cells. This kind of hybrid system has the flexibilityto be optimally sized for different applications so that the advantages of each device can beutilized to maximum extent. At the IEEE Vehicular Power and Propulsion Conference heldon 6–8 September 2006 in Windsor, UK, faculty members at the University of Manchesterpresented their hybrid fuel cell–battery electric London Taxi project with a 6 kW PEMFCand ZEBRA Z5C traction battery [30]. Their results show that the hybridization of a fuelcell and battery can extend the driving range and reduce fuel cell size under dynamicurban driving cycles.

There are several configurations for the design of hybrid systems. In the simplesttopology, the output of the fuel cell stack is connected through a DC–DC converter toa DC link, where fuel cells can be hybridized with other energy storage devices such asbatteries or ultracapacitors. The load is also connected in parallel at the DC link. Notethat the load can be a DC traction motor or an inverter-driven AC traction motor in


Hybrid Controller

DC-DCConverter

DC Link

Battery

Load

PEMHydrogenFuel Cell

Figure 11.28 A simple hybrid fuel cell system

Hybrid Controller

DC-DCConverter

DC Link

Battery

LoadPEM

HydrogenFuel Cell

Ultracapacitor

Figure 11.29 A hybrid fuel cell system with both batteries and ultracapacitors

vehicle applications. Such a topology is shown in Figure 11.28. One key component isthe hybrid controller that will control the power flow among the different power sourcesunder different load conditions. The hybrid controller will be responsible for the hybridsystem’s power and energy management.

The use of batteries can restrict the DC link voltage. Depending on the number of cellsin series, batteries have a definite voltage value. As a result, the output of the convertermust maintain a value corresponding to the battery voltage. However, with the use ofultracapacitors, which are not voltage dependent, the output of the converter can be setto whatever voltage is needed by a particular load. If the use of batteries is desired at theoutput of the converter, ultracapacitors can still be used at the input of the converter orthe output of the fuel cell to further stabilize the system and reduce current spikes in thefuel cell. This hybrid topology is shown in Figure 11.29.

We can also use a double input DC–DC converter (DIC) to accommodate a PEMFCand a battery unit. This hybrid system is illustrated in Figure 11.30. The PEMFC andthe battery unit are connected in parallel so that both sources can supply the load. Thistopology will allow each power source to operate independently with the benefits of eachfully utilized in an appropriate hybrid control strategy. For example, the battery unit canbe voltage controlled to maintain constant DC link voltage VDC , while the PEMFC iscurrent controlled in order to provide the connected load with the required current. TheDIC can be based on the buck–boost converter, which allows the voltage levels of eachpower source to be above or below the DC link voltage level. The battery in Figure 11.30


Hybrid Controller

ACMotor

VFC

VBAT

Dou

ble

Inpu

t DC

/DC

Con

vert

er

VDC+_

3 ϕ

Vol

tage

Sou

rce

Inve

rter

Figure 11.30 A hybrid fuel cell–battery system using a double-input DC–DC converter

PEMHydrogenFuel Cell

Ultracapacitor

Battery

Multi-InputDC–DC

ConverterLoad

Figure 11.31 Hybrid system with a multi-input DC–DC converter

can be replaced by an ultracapacitor to achieve another hybrid fuel cell–ultracapacitorsystem, but a different control strategy will be needed due to the different characteristicsof the battery and ultracapacitor.

Since the buck–boost converters are connected in parallel, the topology can be extendedby connecting additional converters allowing the use of additional sources (i.e., super-capacitor and/or IC engines). Such a topology is illustrated in Figure 11.31, in which amultiple input DC-DC converter is used to allow connections of fuel cell, battery, andultracapacitor to the load. Again, the load can be a DC load or inverter-driven AC load.

A variation of the topology in Figure 11.31 is shown in Figure 11.32, where multipleconverters can be used. The fuel cell would have either a buck or boost converter and thebatteries or ultracapacitors would have a bidirectional buck–boost converter. BidirectionalDC–DC converters allow currents to flow into or out of the energy storage devices andthus allow the devices to supply power to the load or to be charged by either the fuel


PEMFC

Battery

Buck or Boost DC–DC

Converter

Buck –BoostBidirectional

DC–DCConverter

Ultracapacitor

DCLink

Load

Figure 11.32 Hybrid fuel cell energy storage system with a bidirectional converter [27]

cell or electric generator. This topology would also allow for a variable DC link voltage.Both topologies in Figures 11.31 and 11.32 require sophisticated hybrid control strategiesfor the power and energy management.

11.9.4 Control Strategy of Hybrid Fuel Cell Power System∗

In this section, a hybrid power source consisting of a PEMFC and a lead acid battery isused as a case study. The PEMFC is used as the main energy source while the batteryis used as the auxiliary part to improve power quality. A DC–DC boost converter isconnected between the fuel cell and the battery to ensure a proper voltage level at theDC bus and also to control the power flow.

The control strategy takes into account the load profile and the battery SOC. In orderto protect the fuel cell, load transients mitigation control is utilized to filter out the peakvalues of the load. The fuel cell is controlled to provide the steady state load current aftermitigation and the battery supplies the peak currents. The SOC is estimated using thebattery’s voltage and current, which are constantly measured.

In case the SOC is below a specified limit, the fuel cell supplies the charging currentto the battery, until the desired SOC is achieved. Additionally, the control strategy alsoimproves fuel cell efficiency, limits its maximum power output, and controls the loadbus voltage, thus ensuring longer lifetime for the fuel cell and providing reliable workingconditions for the load.

The configuration of the hybrid PEMFC–battery system is shown in Figure 11.33. ThePEMFC, battery, and load are connected in parallel so that both power sources can supplythe load. The PEMFC voltage varies over a large range, depending on the load current.In order to maintain a constant voltage at the DC bus, the fuel cell is isolated from thebattery and the load using a DC–DC boost converter. Using the measured load current,battery current, and DC bus voltage, the main controller calculates the reference currentI ref , which together with the converter output current i con_out is used to determine theduty ratio for the DC–DC converter via the current controller. A three-step-starting DCmotor with a rated voltage of 220 V and a rated power of 2 kW is used as a load.

∗ © [2009] IEEE. Reprinted, with permission.


Load

MainController

CurrentController

PWM pulseGenerator

icon_inicon_out

ibat

Iref

Vcon_outVcon_in

Vbus IBatMea

Iload

A

A

A

V

Battery

DC

DC

FC

+

−

+

−

Icon_out

Figure 11.33 Configuration of the hybrid fuel cell–battery system

The main objective of the hybrid control strategy is to use the fuel cell to satisfy theenergy demands of the load, while the transient peak power demand is covered by thebattery. In order to improve the lifetime of the battery and to maintain the DC bus voltagewithin a small range (±5%) around 220 V, the battery’s SOC is controlled as well. Incase of a low SOC, the fuel cell will provide a charging current.

Two controllers are required for the hybrid control strategy: a current controller and amain controller. The current controller, a PI controller, regulates the output current i con_out

of the DC–DC converter by controlling the duty ratio of the PWM generator. Referencecurrent I ref is provided by the main controller. The tasks of the main controller consistof regulating the steady energy for the load and any additional charging energy for thebattery, both provided by the PEMFC. Thus, the main controller is split into two parts:the load transient mitigation controller [31], which supplies steady load current i LoadREF ;and the battery SOC controller, which supplies the battery charging reference currenti BatREF . Details of the two controllers can be found in [32]. The flow chart of the overallcontrol strategy is shown in Figure 11.34. The objective of the control strategy is toprovide an optimum response to the load demand, to maintain the SOC of the battery(i.e., maintaining the energy reserve), and to ease operation of the PEMFC (i.e., extendingthe life cycle of the fuel cell).

Two different simulation scenarios are presented. The first scenario includes a lead acidbattery pack, consisting of 17 cells connected in series, with a relatively high capacity of1 Ah for each battery cell up to a total of 17 Ah for the battery stack. With such a highcapacity the battery stack should be able to provide the load with enough peak powerwithout needing to be charged. The second scenario includes a battery pack, with thesame amount of batteries, with a significantly lower capacity of 0.05 Ah each up to a totalof 0.85 Ah for the battery stack. The batteries in this scenario should be discharged to apoint where charging will become necessary.


iBat REF = 0

iLoadREF = iLoad min

iREF = iBatREF + iLoadREF

iLoadREF < iLoad min

iBatchar ≥ iBat max

iBatREF ≥ iBatchar

iBatchar = iBat max

iBatREF = iBatchar

iREF = iREF maxiREF > iREF max

output iREF

ΔVbus

Vbus max –VbusibatREF = iBatchar ×

or iLoadREF = iLoad after filter

tiLoadREF =

∫ iload dt

Start

Read load currentRead battery current

Read bus voltage

SOC<85%SOCREF

SOC>=SOCREF

YesMode=1

Mode=0

Mode=1

Yes

Yes

Yes

Yes

Yes

Yes

No

No

No

No

No

No

Calculate battery SOC

IBat maxiBatchar = (SOCREF – SOC) ×

ΔSOC

Figure 11.34 Flow chart of the hybrid control strategy


The purpose of using two different battery capacities is to illustrate the battery chargingcontrol algorithm implemented in the system, where the fuel cell is used not only to supplythe load demand, but also to maintain the SOC of the battery.

It is assumed that at the beginning of the simulations the batteries are fully chargedwith the SOC equal to 1.

11.9.4.1 Maximum Capacity of 1 Ah per Battery Cell

As expected, the SOC of the battery stack with such a high capacity remains above 85%.Thus no charging current provided by the fuel cell is necessary in this scenario. There-fore, the main controller output i ref equals the converter output current i con_out , whichis shown in Figure 11.35. The low-pass filter, implemented in the load mitigation con-troller, smoothes the transients of the load current providing a steadily growing referencecurrent, resulting in a smooth converter output current. At the same time, the batterycovers all peak power transients of the load and is zero at steady state load, as shown inFigure 11.36.

The fuel cell output current as shown in Figure 11.37 has the same profile as theconverter output current. As can be seen at the starting point, the fuel cell is controlled tohave a rapidly increasing output current to avoid low efficiency due to small output current.

As an auxiliary part of the fuel cell system, the battery may not need a very high capac-ity. In comparison the fuel cell is supposed to be a large energy storage source. Thereforea simulation for a smaller battery capacity (0.05 Ah per battery cell) was performed.

11.9.4.2 Maximum Capacity of 0.05 Ah per Battery Cell

The output current of the fuel cell shows a smooth profile, as seen from the inset inFigure 11.38, until the third transient of the load at around five seconds of the simulation,when the battery SOC drops below 85% and the charging of the battery is triggered. Thefuel cell is controlled to provide charging current until the battery is fully charged again,

Reference Current

Time (s)0 2.5 5 7.5 10 12.5 15 17.5 20

Cur

rent

(A

)

0

2.5

5.0

7.5

10.0 I_REFI_CON

Figure 11.35 Reference current and converter output current (1 Ah/battery cell)


Battery Current

Time (s)0 2.5 5 7.5 10 12.5 15 17.5 20

Cur

rent

(A)

0

10

20

30

40

Figure 11.36 Battery output current (1 Ah/battery cell)

Fuel Cell Current

Time (s)

0 2.5 5 7.5 10 12.5 15 17.5 20

Cur

rent

(A

)

0

2

4

6

8

10

12

14

16

Figure 11.37 Fuel cell output current (1 Ah/battery cell)

Fuel Cell Current

Time (s)0 2.5 5 7.5 10 12.5 15 17.5 20

Cur

rent

(A

)

0

2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

Fuel Cell Current

Time (s)4.5 5.5 6.5C

urre

nt (A

)

10

15

20

Figure 11.38 Fuel cell output current (0.05 Ah/battery cell)


Battery State of Charge

Time (s)

0 2.5 5 7.5 10 12.5 15 17.5 20

SOC

(%

/100

)

.850

.875

.900

.925

.950

.975

1.000

Figure 11.39 Battery state of charge (0.05 Ah/battery cell)

as shown in Figure 11.39. A dead-band is necessary to avoid the chattering effect of thecontroller from charging/discharging the battery at a faster rate. At the beginning of thecharging mode, the current output of the fuel cell significantly steps up, following theincrease of the reference current from the current controller, providing charging currentto the battery and mitigated load current to the DC motor. However, the increase in thefuel cell current seems to be smooth enough due to the low-pass filter action in the loadcurrent measurement, thus assuring a longer life cycle for the fuel cell in applicationswith dynamic load profiles.

The simulation results show that the whole system works well and both the fuel cellcurrent and battery current can be controlled as desired to meet different load demands.

11.10 Summary and Discussion

In this chapter, the most important energy storage devices – batteries, ultracapacitors,flywheel energy storage systems, hydraulic energy storage systems, and fuel cells – arediscussed in detail. Since many cells of batteries and ultracapacitors are connected inseries and parallel to achieve desired voltage and current ratings, equalizing circuits areindispensible for the energy management and control of onboard energy storage devices.Different battery charging techniques that are commonly used for charging the battery inHEVs are compared. There are different advantages and disadvantages for each charg-ing method. Conventional methods like CC and CV have many disadvantages; the pulsecharging technique is gaining popularity and will be a better option for HEV batterycharging. Integration of an ultracapacitor with a battery can potentially mitigate the dis-advantages of each component and deliver superior performance for a HEV or PHEV.Fuel cell technologies are promising for future clean vehicles, but current PEMFCs cannotmeet the requirements of high peak power, low cost, and robustness. Therefore, hybridiza-tion of fuel cells with batteries or ultracapacitors will result in higher energy efficiencywith regenerative braking capability under dynamic driving cycles.


References1. Zheng, J.P., Jow, T.R., and Ding, M.S. (2001) Hybrid power sources for pulsed current applications. IEEE

Transactions on Aerospace and Electronic Systems , 37 (1), 288–291.2. Van Mierlo, J., Van den Bossche, P., and Maggetto, G. (2004) Models of energy sources for EV and

HEV: fuel cells, batteries, ultracapacitors, flywheels and engine-generators. Journal of Power Sources ,128, 76–89.

3. Brooker, A., Haraldsson, K., Hendricks, T. et al. (2002) ADVISOR Documentation (Version 2002),National Renewable Energy Laboratory.

4. Brian, M. (2006) How Lithium-ion Batteries Work. HowStuffWorks.com. http://electronics.howstuffworks.com/lithium-ion-battery.htm (accessed November 14, 2006).

5. Lee, Y. and Cheng, M. (2005) Intelligent control battery equalization for series connected lithium-ionbattery strings. IEEE Transactions on Industrial Electronics , 52 (5), 1297–1307.

6. Affanni, A., Bellini, A., Franceschini, G. et al. (2005) Battery choice and management for new-generationelectric vehicles. IEEE Transactions on Industrial Electronics , 52 (5), 1343–1349.

7. Maxwell Technologies, (2006) Fuel Cells and Ultracapacitors – A Proven Value Proposition Versus Incum-bent Technologies, San Diego.

8. Maxwell Technologies, (2011) Ultracapacitors Help P21 to Provide Fuel Cell-Based Backup Power forTelecoms, San Diego.

9. Larminie, J. and Lowry, J. (2003) Electric Vehicle Technology Explained , John Wiley & Sons, Ltd, Chich-ester.

10. Dougal, R.A., Gao, L., and Liu, S. (2004) Ultracapacitor model with automatic order selection and capacityscaling for dynamic system simulation. Journal of Power Sources , 126, 250–257.

11. Sule, V. and Santoso, S. (n.d.) Constant Current and Voltage Battery Charging Schemes for Stand-AloneWind Turbines, Department of Electrical and Computer Engineering, University of Texas.

12. Jiang, Z. and Dougal, R.A. (2004) Synergetic control of power converters for pulse current chargingof advanced batteries from a fuel cell power source. IEEE Transactions on Power Electronics , 19 (4),1140–1150.

13. Nemeth, M.S. (2002) The 12 volt Side of Life, http://www.ccis.com/home/mnemeth/12volt/12volt.htm(accessed March 3, 2002).

14. Park, H., Kim, C.-E., Kim, C.-H. et al. (2009) A modularized charge equalizer for an HEV lithium-ionbattery string. IEEE Transactions on Industrial Electronics , 56 (5), 1464–1476.

15. Kutkut, N., Wiegman, H.L.N., Divan, D., and Novotny, D. (1998) Charge equalization for an electricvehicle battery system. IEEE Transactions on Aerospace and Electronic Systems , 34 (1), 235–245.

16. Moo, C., Hsieh, Y., and Tsai, I. (2003) Charge equalization for series-connected batteries. IEEE Transac-tions on Aerospace and Electronic Systems , 39 (2), 704–710.

17. Kutkut, N.H., Wiegman, H.L.N., Divan, D., and Novotny, D. (1999) Design considerations for chargeequalization of an electric vehicle battery system. IEEE Transactions on Industry Electronics , 35 (1),28–35.

18. Miller, J.M. (2004) Propulsion Systems for Hybrid Vehicles , IEE, London.19. Baughman, A. and Ferdowsi, M. (2008) Double-tiered switched-capacitor battery charge equalization tech-

nique. IEEE Transactions on Industrial Electronics , 55 (6), 2277–2285.20. Chan, C.C. and Chau, K.T. (2001) Modern Electric Vehicle Technology , Oxford University Press, Oxford.21. Jefferson, C.M. and Barnard, R.H. (2002) Hybrid Vehicle Propulsion , WIT Press, Boston, MA.22. Markel, T., Zolot, M., Wipke, K.B., and Pesaran, A.A. (2003) Energy storage requirements for hybrid fuel

cell vehicles. Advanced Automotive Battery Conference, June 10–13, Nice, France.23. Matsumoto, T., Watanabe, N., Sugiura, H., and Ishikawa, T. (2002) Development of fuel-cell hybrid vehicle.

SAE World Congress, March 4–7, Detroit, MI, paper 2002-01-0096.24. Rodata, P., Garcia, O., Guzzella, L. et al. (2003) Performance and Operational Characteristics of a Hybrid

Vehicle Powered by Fuel Cells and Supercapacitors, SAE paper 2003-01-0418, Society of AutomotiveEngineers.

25. Wipke, K.B., Markel, T., and Nelson, D. (2001) Optimizing energy management strategy and degree ofhybridization for a hydrogen fuel cell SUV. 18th International Electric Vehicle Symposium (EVS 18),October, Berlin, Germany.


26. Larminie, J. and Dicks, A. (2000) Fuel Cell Systems Explained , John Wiley & Sons, Ltd, Chichester.27. O’Hayre, R., Cha, S-W., Colella, W., and Prinz, F.B. (2005) Fuel Cell – Fundamentals, John Wiley &

Sons, Inc., New York.28. Wang, C., Nehrir, M.H., and Gao, H. (2006) Control of PEM fuel cell distributed generation systems. IEEE

Transactions on Energy Conversion , 21 (2), 586–595.29. Wang, C. and Nehrir, M.N. (2003) A dynamic model for PEM fuel cells using electrical circuit. Proceedings

of 35th North American Power Symposium, October, Rolla, MO, pp. 30–35.30. Lachichi, A. and Schofield, N. (2006) Comparison of DC-DC converter interfaces for fuel cells in electric

vehicle applications. Proceedings of the IEEE Vehicle Power and Propulsion Conference, September 6–8,Windsor, UK.

31. Wang, C. and Nehrir, M.H. (2007) Load transient mitigation for stand-alone fuel cell power generationsystems. IEEE Transactions on Energy Conversion , 22 (4), 864–872.

32. Zheglov, V., Gao, W., Muljadi, E., and Wang, G. (2009) New control strategy for stand-alone fuel cell-battery hybrid power supply system. IEEE PES General Meeting, July, Calgary, Canada.

Documents

Hybrid Electric Vehicles (Principles and Applications with Practical Perspectives) || Batteries, Ultracapacitors, Fuel Cells, and Controls